vgarvardt/img.py

## img.py
#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# Video of original screencast: https://vimeo.com/57296525
#
# Modified by Vladimir Garvardt
#
from __future__ import print_function

import os
import random

from PIL import Image as pImage
import numpy

class Image:
    """Take an information from image file"""

    BLOCK_SIZE = 20
    TRESHOLD = 60
    MAX_DISTANCE = 220

    def __init__(self, filename):
        self.filename = filename
        self.checked = False
        self.load()

    def load(self):
        img = pImage.open(self.filename)
        img = img.resize((Image.BLOCK_SIZE, Image.BLOCK_SIZE), pImage.BILINEAR)
        self.t_data = numpy.array(map(sum, img.getdata()))

    def similar(self, other):
        if self.checked: return None
        if self == other: return -1

        dist = sum(1 for x in self.t_data - other.t_data if abs(x) > Image.TRESHOLD)
        return None if dist > Image.MAX_DISTANCE else dist

    def __repr__(self):
        return self.filename

    def __eq__(self, obj):
        return self.filename == obj.filename


def get_images_list(dirname):
    files = (f for f in os.listdir(dirname) if f.lower().endswith('.jpg'))
    files = (os.path.join(dirname, f) for f in files)
    images = [Image(f) for f in files]
    random.shuffle(images)
    return images

def find_similar(images):
    similar = []
    for img in images:
        lookin = filter(lambda x: not x.checked, [x for x in images])
        distances = filter(lambda d: d[0] is not None, [(img.similar(x), x) for x in lookin])
        if len(distances) > 1: similar.append(distances)
        img.checked = True
    return similar

def htmlize_similar(similar):
    res = '<html><body>\n'
    for distances in similar:
        for dist, img in distances:
            res += '<img src="%s" width="200"> %s\n' % (os.path.basename(img.filename), dist if dist > -1 else '-')
        res += '<hr/>\n'
    res += '</body></html>'
    return res


if __name__ == '__main__':
    images = get_images_list('./') # './,~/Pictures'
    similar = find_similar(images)
    print(htmlize_similar(similar))
	#!/usr/bin/env python
	# -- coding: utf-8 --
	#
	# Video of original screencast: https://vimeo.com/57296525
	#
	# Modified by Vladimir Garvardt
	#
	from __future__ import print_function

	import os
	import random

	from PIL import Image as pImage
	import numpy

	class Image:
	"""Take an information from image file"""

	BLOCK_SIZE = 20
	TRESHOLD = 60
	MAX_DISTANCE = 220

	def __init__(self, filename):
	self.filename = filename
	self.checked = False
	self.load()

	def load(self):
	img = pImage.open(self.filename)
	img = img.resize((Image.BLOCK_SIZE, Image.BLOCK_SIZE), pImage.BILINEAR)
	self.t_data = numpy.array(map(sum, img.getdata()))

	def similar(self, other):
	if self.checked: return None
	if self == other: return -1

	dist = sum(1 for x in self.t_data - other.t_data if abs(x) > Image.TRESHOLD)
	return None if dist > Image.MAX_DISTANCE else dist

	def __repr__(self):
	return self.filename

	def __eq__(self, obj):
	return self.filename == obj.filename


	def get_images_list(dirname):
	files = (f for f in os.listdir(dirname) if f.lower().endswith('.jpg'))
	files = (os.path.join(dirname, f) for f in files)
	images = [Image(f) for f in files]
	random.shuffle(images)
	return images

	def find_similar(images):
	similar = []
	for img in images:
	lookin = filter(lambda x: not x.checked, [x for x in images])
	distances = filter(lambda d: d[0] is not None, [(img.similar(x), x) for x in lookin])
	if len(distances) > 1: similar.append(distances)
	img.checked = True
	return similar

	def htmlize_similar(similar):
	res = '<html><body>\n'
	for distances in similar:
	for dist, img in distances:
	res += '<img src="%s" width="200"> %s\n' % (os.path.basename(img.filename), dist if dist > -1 else '-')
	res += '<hr/>\n'
	res += '</body></html>'
	return res


	if __name__ == '__main__':
	images = get_images_list('./') # './,~/Pictures'
	similar = find_similar(images)
	print(htmlize_similar(similar))